Madhu Pattarambil is not a typical software founder. He has spent 25 years in chemical engineering, hazardous waste management and process safety, and still works full-time as a national dangerous goods specialist at REMONDIS in Sydney. The idea for Madram Milsan AI came from a single frustrating Sunday: trying to plan a family holiday, he ended up with eleven browser tabs open across flights, hotels, visas, insurance, eSIMs, weather and food, and two hours later had only fragments of a plan. He closed the tabs and opened a code editor instead.

Two weeks later the app was live. Madram Milsan AI takes any destination and produces a complete trip plan: a day-by-day schedule, a local food guide, clothing suggestions tied to real weather, a full budget breakdown, plus live flight, hotel, insurance, forex and eSIM options. It covers 140 destinations across more than 80 countries and ships in 17 languages from day one. There is no account required and no paywall.

The build itself was deliberately plain. Madhu used the Claude API for the planning intelligence, vanilla JavaScript for the frontend, two serverless functions, and Netlify for deployment, with no framework or build step. Delivered as a progressive web app, it installs straight from the browser, which sidesteps app-store approvals and platform fees. His argument is that 25 years of understanding how complex systems fail mattered more than a computer science degree.

The money side is built around affiliate commissions rather than subscriptions or ads. When a traveller books a flight, hotel, car, eSIM or insurance through the app, partners such as Aviasales, Booking.com, DiscoverCars, Wise, Airalo, Holafly, Klook and KKday pay a commission at no extra cost to the user. Madhu frames this as the only model where his incentives match the traveller's: he earns only when someone finds something genuinely useful and books it.

A note on the numbers: this story is based on Madhu's own launch write-up published in mid-June 2026, days before this entry was added. He has not disclosed any revenue, booking volume or traffic figures, so none are claimed here. What is documented is the build itself and the affiliate model behind it. The product name encodes his family, Madhu, Ramya, Milindh and Sandra, which he says keeps him accountable to ship something he would be proud to show them.